Implementation science theories, models and frameworks (TMFs) can guide implementation research and practice to address widening inequities in healthcare access, quality, and health outcomes. This scoping review aimed to map and critically appraise implementation science TMFs explicitly created or adapted to advance health equity. We conducted a scoping review using Joanna Briggs Institute guidance. We searched MEDLINE, CINHAL, and five implementation science and equity journals from inception to May 20, 2025. Eligible records were primary descriptions of (i) implementation science TMFs purposely built to advance health equity or (ii) equity-focused adaptations of existing implementation science TMFs. Applications of TMFs in research or practice and non-implementation science TMFs were excluded. Two reviewers independently screened titles/abstracts then full texts in Covidence with consensus adjudication. Data were charted on a standardized form, and each TMF underwent duplicate critical equity appraisal against eight a priori criteria with an equity expert providing methodological input. Findings were summarized narratively and in evidence tables. From 6,012 records, 42 TMFs were included (23 adaptations of existing TMFs, 19 purpose-built), and most (69
Audit Feedback (A F) is a common intervention to enhance clinical guideline adherence, yet uncertainty remains regarding the factors influencing its effectiveness. To examine the effectiveness of A F interventions targeting clinical guideline adherence of hospital-based medical specialists in general and to determine which design characteristics, co-interventions and contextual factors explain variance in A F effectiveness. A systematic review and meta-analysis of randomized controlled trials was conducted following the Cochrane Handbook methodology. The electronic databases MEDLINE, Embase and Web of Science were searched up to January 2024. Two reviewers independently screened articles and selected studies on A F interventions targeting clinical guideline adherence of hospital-based medical specialists. Risk of bias was assessed by two independent reviewers, following the ROB2 tools. Data on design characteristics, co-interventions, contextual factors and outcomes were extracted by one reviewer and checked by a second reviewer. Results were described descriptively and when appropriate meta-analysis and subgroup analyses using random effects modelling was performed. Forty-one studies were included. Studies had a 2-armed (N = 35), 3-armed (N = 5) or 2 × 2 (N = 1) design and compared A F vs. no intervention (N = 15), A F vs. another intervention (N = 8) or A F vs. A F (N = 20). Studies targeted 23 different medical specialties, of which Cardiology (N = 8), General Medicine (N = 7), ICU (N = 6) and Emergency Care (N = 6) were most targeted. Risk of bias was low in six studies, with some risk in 19 studies and high risk in 16 studies. Meta-analysis showed a significant improvement of clinical guideline adherence due to A F, with a pooled OR of 1.30 (95
Artificial intelligence (AI) encompasses computational systems that perform tasks typically requiring human intelligence, including machine learning, generative AI, and agentic applications. The use of AI to support implementation activities is growing, but its applications and evaluation remain poorly characterised. We conducted the first cycle of a living scoping review to identify how AI is being used and evaluated across implementation science and practice. We followed JBI and Cochrane guidance and reported findings per PRISMA-ScR and PRISMA-LSR. We searched six databases through April 6, 2026, and included sources describing or evaluating AI to support implementation research or practice activities. We excluded adjacent uses such as knowledge synthesis automation. We extracted study characteristics, AI approaches, implementation tasks, evaluation methods, outcomes, and risks, and synthesised findings descriptively. We identified 7,203 records and included 40 sources, 34 (85
Digital health technologies (DHTs) have proven to be valuable tools in managing chronic conditions. However, significant barriers remain in the implementation of DHTs. A gap remains in understanding the comparative effectiveness of implementation strategies in the context of digital realm. This study aims to make a comprehensive evaluation of the implementation strategies while also investigating potential moderators influencing strategy effectiveness. This study is a systematic review and meta-analysis. This study searched four databases and hand searched reference lists from inception to July 2024 for randomized and non-randomized trials with a control group and interrupted time series assessing the impact of at least one implementation strategy to improve DHTs implementation using an outcome that could be mapped to the implementation, patient, and service outcomes based on the Implementation Outcome Framework. Random-effects meta-analysis was conducted. We also conducted meta-regression and subgroup analysis to explore the moderators of the strategy effects and heterogeneity. This study was registered with PROSPERO (CRD42025519493). We screened 9544 records and selected 32 articles for systematic review and 24 articles were included in the meta-analysis. The most frequently used strategies are “engage consumers”, “support clinicians”, and “train and educate stakeholders”. Most of the articles were with high or serious risk of bias. Implementation strategy as a whole showed a small positive average effect on implementation outcomes compared with no implementation strategy (RD = 0.02, 95
Implementation science is moving from a proliferation of theories, models, and frameworks toward more rigorous empirical methods. However, quantitative implementation data remain highly heterogeneous, with non-standardized instruments, variable operationalizations, and context-specific adaptations that challenge comparison, replication, and cumulative knowledge building. While data harmonization has advanced in other disciplines, implementation science has yet to systematically adopt or adapt these methods, despite growing needs for cross-study synthesis. This Methodology paper addresses the challenges of harmonizing quantitative implementation data across multiple studies. We outline categories of implementation research questions that would benefit from harmonization, characterize the complex nature of implementation data, and identify technical, conceptual, and analytic challenges important to the field. We then propose methodological strategies and solutions to address data harmonization issues in implementation science. Core challenges of harmonizing implementation data were identified: (1) construct validity; (2) measurement alignment; (3) psychometric gaps; (4) contextual heterogeneity; and (5) temporal differences in data collection. At the same time, harmonization offers substantial gains, including increased statistical power, the ability to test mediators and moderators, improved generalizability, and expanded capacity for mechanism-focused modeling. Potential solutions to these issues include the use of calibration datasets with multiple imputation and inverse probability weighting, latent variable and growth models to link non-equivalent instruments, time-to-event and stage-based approaches to address temporal misalignment, and the development of unified yet adaptable instruments and common data elements to support coordinated data collection across studies. Data harmonization represents a critical pathway for advancing cumulative science in implementation research. Yet harmonization alone cannot resolve underlying conceptual inconsistencies; progress will require clearer construct boundaries, strengthened psychometrics, and flexible strategies that balance comparability with preservation of contextual meaning. Future efforts will benefit from implementation-specific harmonization methods, principled decisions about when harmonization is appropriate, and collaborative infrastructures to support data sharing and standardization across studies and settings.
Delirium affects up to two-thirds of palliative care unit (PCU) inpatients, causing distress for patients, families, and staff, and contributing to complex care needs and increased healthcare resource use. Although evidence-based guidelines exist, their consistent implementation in PCUs is limited, and implementation strategies to support guideline-adherent care have not been rigorously evaluated in this setting. Creating Learning Environments for Compassionate Care-Palliative Delirium (CLECC-Pal Delirium) is a co-designed multi-component implementation strategy to help embed guideline-adherent delirium care within routine PCU practice. This cluster randomised controlled trial (cRCT) evaluates the effectiveness and cost-effectiveness of the implementation strategy using an implementation-to-target design, with embedded economic and process evaluations. Adaptive implementation-to-target, Type III hybrid effectiveness-implementation parallel group cRCT across 20 PCUs (10 per arm, 50 patient records per cluster at each timepoint) in the UK, randomised 1:1 to CLECC-Pal Delirium or usual care. Randomisation is stratified by unit size, education provision, and provider type (NHS/Charity). Sites will receive tailored support to achieve predefined implementation-readiness criteria prior to data collection. The primary outcome is the proportion of each admission’s inpatient days affected by delirium (delirium days), measured retrospectively using a validated chart-based method adapted for palliative care. Secondary outcomes include adherence to delirium care guidelines, patient symptom burden, functional status, and incremental cost-effectiveness. The sample size (1200 admission episodes with delirium) provides 92.3
Practice facilitation is an effective multicomponent strategy that focuses on building primary care practice capacity for continuous quality improvement. Despite its growing use, few frameworks empirically specify how practice facilitation strategies are operationalized in resource-constrained settings such as small, independent primary care practices (SIPs). This study aimed to further expand our understanding of core practice facilitation processes and strategies used to facilitate the adoption of evidence-based interventions in SIPs. We conducted a qualitative analysis of practice facilitator (PF) field notes from a stepped-wedge randomized controlled trial evaluating the impact of practice facilitation on adoption of team-based care (TBC) for hypertension management in 74 SIPs across New York City. Facilitators conducted 16 structured site visits per practice over a 12-month intervention period and documented implementation activities after each visit using a standardized digital platform. We analyzed 528 visit transcripts using a hybrid deductive–inductive approach. Deductive coding focused on established practice facilitation strategy domains; inductive coding identified novel processes and strategies that emerged from the data. Coding was conducted iteratively with regular team calibration and member checking with the PF team. We identified an iterative process in which PFs observe/listen, diagnose, engage, and problem solve throughout the implementation period – a dynamic approach not described in previous frameworks. This process enhanced PFs’ understanding of each practice’s context and guided their selection and application of a core set of practice facilitation strategies. These included previously documented strategies such as training, coaching, educating, and facilitating, as well as a novel strategy we identified: modeling, in which PFs demonstrated specific tasks or workflows in real-time alongside staff, building confidence and reinforcing learning. This study refines and extends current facilitation frameworks to further define core processes and strategies that drive practice transformation in SIPs. The framework provides practical guidance for implementation efforts targeting small practice settings. ClinicalTrials.gov; NCT05413252; 2022-06-09.
The treatment of depression is essential for improving both psychiatric and medical outcomes for youth with HIV (YWH). Previous studies have shown that the combination of a medication algorithm and CBT tailored for YWH (COMB) is efficacious for decreasing depressive symptoms and improving quality of life, but sustainability of effects remains difficult. Most recently, a cluster-randomized RCT found at Week 24 statistically significant improvement in the site-level mean number of depressive symptoms, the proportion of YWH with a treatment response, and the proportion in remission at COMB sites compared to treatment as usual (TAU) sites (6.7 vs. 10.6, 62
Nurse-initiated care improves timely treatment access, but adoption varies. The Ministry of Health introduced 73 Emergency Care Assessment and Treatment (ECAT) protocols for public emergency departments (EDs) to standardise nurse-initiated care. A behaviour change strategy was used to implement ECAT in 29 EDs as part of a trial, using implementation strategies like education, clinical champions, videos, and audit and feedback. The aim of this study was to evaluate the implementation of the ECAT protocol and the strategies used for reach, effectiveness, adoption, quality, and maintenance (REAIM). This multi-method implementation evaluation used the RE-AIM framework. Methods comprised: i) emergency nurse surveys (Intervention and Control groups); ii) scoring by site implementation nurses of fidelity, adaptation and effectiveness of implementation strategies out of 5; iii) audits determining appropriateness of ECAT protocol use at 6–12 weeks and iv) implementation tracking logs during all study periods. Descriptive statistics were used for quantitative data, and content analysis for qualitative data (surveys/logs). Data from surveys (n = 787), audits (1375 audits, 509 nurses), fidelity scores (n = 11), and implementation tracking logs (>600 entries) were used to evaluate implementation of ECAT protocols in the 29 EDs. Emergency nurses reported high use of ECAT protocols in their daily practice (median (IQR) 9.0 (8.0, 10.0), and audits demonstrated 92.9
Qualitative methods are now central to implementation science, but they are used to do different kinds of work that are not always clearly distinguished. Some studies are designed to develop contextual or conceptual understanding, while others are expected to inform near-term decisions about rollout, adaptation, or implementation strategies. When these differences remain implicit, qualitative studies may be designed and evaluated against expectations they were never intended to meet, particularly around the role of theory, openness to unanticipated findings, and what counts as rigor. In this debate paper, we argue for a more explicit way of thinking about qualitative inquiry in implementation science. Rather than treating qualitative methods as a single approach, we suggest they are being configured in different ways in response to study purpose, time and resource constraints, the state of knowledge about the phenomenon or context, and stakeholder expectations. To make these differences visible, we propose a positioning framework that locates qualitative inquiry along a continuum of three orientations: Generative, Pragmatic, and Action-oriented. Generative inquiry prioritizes contextual depth and conceptual development; Action-oriented inquiry is organized to produce timely, decision-relevant findings; Pragmatic inquiry occupies the space between these poles. We then introduce a positioning guide and an accompanying table to show how these orientations shape key aspects of qualitative design, including question formulation, the role of theory, sampling, data collection, analysis, and reporting. The contribution of this paper is a framework for describing what qualitative studies in implementation science are trying to produce and how they should be assessed. Qualitative rigor cannot be reduced to a single standard when studies are making different kinds of claims. Making methodological positioning more explicit may help reduce mismatched expectations in study design and peer review, and support more consistent judgments about the contribution of qualitative work in implementation science.
Dissemination science lacks shared language for specifying what dissemination strategies are designed to achieve. Without clearly defined outcomes, researchers cannot design studies that explain why dissemination strategies succeed or fail, compare findings, or build cumulative evidence about how dissemination works. We drew on dissemination scholarship, communication and behavior theory, innovation diffusion, public policy, and organizational readiness frameworks, and two decades of applied experience within a national HIV research network, to identify and iteratively refine a set of dissemination outcomes. We propose a taxonomy of eight dissemination outcomes: Exposure, Comprehension, Credibility, Salience, Perceived Fit, Leadership Endorsement, Action Readiness, and Decision to Implement. We distinguish these from dissemination mechanisms and from implementation outcomes, locating the boundary between dissemination and implementation at the decision to implement. We illustrate the taxonomy using three studies from the Adolescent Medicine Trials Network for HIV Interventions and offer it not as a prescriptive framework but as a conceptual starting point. We hope this shared language supports clearer study design and helps build cumulative evidence about how dissemination works.
